verify: strip white-space from manifest before comparing

[Freax13]

Previously we failed verification when the white-space between the two manifest files didn't match.

[burgerdev]

If we wanted to introduce canonicalization, I would rather use e.g. RFC 8785, but we'd need to use that everywhere the manifest is interpreted.

However, I don't understand the use case:

1. The workload owner has the manifest sitting in their workspace.
2. The data owner should have received the expected manifest from the workload owner.

In both cases, I don't see why the manifest would have changed in whitespace.

We had a discussion about a similar proposal a few weeks back: #615 (comment).

[msanft]

In both cases, I don't see why the manifest would have changed in whitespace.

Without knowing any other context, I can think of a scenario where one editor appends a newline at the end, or uses spaces for tabs, and the other doesn't.

I generally think comparisons of structured data should happen in a canonical format, but wold also favor RFC-canonicalization over stripping.

[Freax13]

Closed for the following reasons:

• There's no demand for this/there's no real use case.
• Comparing the manifests byte-for-byte is very simple and therefor secure. Canonicalizations can potentially be a source for bugs, so we should avoid them unless there's a good reason.